摘要
本文在传统的尺度不变特征子(Scale Invariant Feature Transform,SIFT)算法的基础上,提出了一种新的基于改进的SIFT压缩感知跟踪算法.该方法一方面通过改进压缩跟踪算法中分类器的更新策略来提高算法的实时性;另一方面,通过改进SIFT向量邻域的选取方法来实现降低向量维度,从而减少计算复杂度.仿真实验表明,该方法不仅可以提高跟踪目标的实时性,而且能够在发生目标尺度变化、遮挡、漂移的情况下对运动目标进行准确跟踪.
An improved SIFT target tracking algorithm based on compressed sensing is proposed in this paper. The real-time performanceis improved by improving updating strategy of the classifier in the com- pressive theory. On the other hand, The vector neighborhood of SIFT has been improved to decrease the vector dimension and the complexity of calculation. Simulations and experiments show that this method not only can improve the real-time performance of tracking target, but also can carry on the tracking of moving target accurately in the event of a target scale variation, occlusion shelter, drifting.
作者
庄哲民
龚家铭
谢光成
袁野
ZHUANG Zhemin GONG Jiaming XIE Guangcheng YUAN Ye(Dept. of Electronic Engineering, Shantou University, Shantou 515063, China)
出处
《测试技术学报》
2017年第2期93-99,共7页
Journal of Test and Measurement Technology
基金
国家自然科学基金资助项目(61471228)
广东省应用型科技研发专项项目(2015B020233018)
关键词
运动目标跟踪
特征提取
SIFT
压缩感知
video target tracking
feature extraction
SIFT
compressive sensing
